The video game business is an expensive and risky hit-driven market. But live-service games are altering this dynamic. Instead of trying to hit a home run with a massive blockbuster, companies are trying to build platforms that people will come back to for months or years.

Ubisoft, for example, has explicitly said that it doesn’t want to make game products anymore — it wants to make platforms. It has seen success doing that with Rainbow Six: Siege, and it has emulated that with For Honor and The Division.

But if you’ve played games for a significant amount of time, you may wonder why this change is happening. Why are publishers like Ubisoft abandoning a product model that worked for decades? Well, you can find the answer to that in the Steam peak concurrent players chart. And YouTube channel TheRankings has visualized that list over the four-year period from 2015 through 2018 to make it easier to parse.

The video shows the peak concurrent players for the top 15 games on Steam every day starting on January 1 2015 and running through the end of last year. As games grow or as new games get released, they jump up the chart. And as people lose interest in existing games, they fall out of the top 15.

You can watch the chart in action by clicking play on the video at the top of this list. But let’s talk about what it highlights and what it reveals.

Blockbusters exist, but they come and then go

The first thing I noticed while watching this video is just how infrequently we get games that are big enough to make a dent on Steam. And even when a blockbuster does come along, it tends to falls off the charts after a few weeks — or sometimes in a few days.

Did you even notice The Witcher III: Wild Hunt on the chart? It popped up to No. 4 when it came out in May 2015. But by July 2015, it was gone.

The Witcher III is one of the biggest games of the last five years. But CD Projekt Red did not build it to hang on this list. It’s a single-player campaign that you experience and then put away. Games like that can obviously still make money. Witcher III certainly has. But the behavior of this chart highlights how a game like The Witcher III may not get the best return on investment.

Live-service games stick around for years and can even grow over time

By January 2015, Warframe was already almost two years old. The free-to-play sci-fi shooter was also still among the 15 most played games on Steam. And as the video reveals, it was consistently in the top 15 all the way through December 2018.

What’s most impressive about Warframe, however, is that its daily peak concurrent number grows over time. On January 1, 2015 it has 24,000 players. That number gets leaps over 100,000 players by October 2017. It does so again in August 2018. Throughout all of last year, Warframe’s daily peak concurrent players is in the 70,000-to-80,000 range with those occasional peaks closer to 100,000.

Warframe started small, and it has grown over time. And it’s not the only game to follow that trajectory.

Ubisoft’s Rainbow Six: Siege did not crack the top 15 when it debuted in December 2015. It doesn’t even consistently rank among the most-played games on Steam until early 2017. But it has turned into a monster ever since. It’s gone from approximately 40,000 daily peak concurrent players on Steam in 2017 to regularly having more than 100,000 by the end of 2018.

Other live service games have seen similar staying power. The car-soccer game Rocket League is almost always in the top 10. And survival sims Rust and Ark: Survival Evolved are also nearly always on the chart.

This doesn’t even count the thriving and growing audience for live-service games on console, which includes Warframe, Rocket League, and Siege.

Is the risk worth the investment

What this chart drives home more than anything is that making a massive, triple-A blockbuster is potentially not worth the cost.

The Witcher III came and went. But even something like Fallout 4 quickly fell back behind games like Rust and Ark. In fact, the one blockbuster that hasn’t lost its place at the top of the chart is Grand Theft Auto V, and that has a massively popular live-service mode in GTA Online.

Game companies don’t share their budgets, but it’s probably fair to assume that The Witcher III is a more expensive game to make than Rocket League.

And that is the crucial point here. Game companies have this data. And they are learning the lesson that even if they make a game as beloved as The Witcher III (which most developers probably cannot do), people are still going to put it away after a few weeks. But if they make a live-service game, even if it’s not perfect at launch and starts small, they have a chance of making something that could turn into a consistent source of revenue. On top of that, a live-service game is probably less expensive to make. And while they are functionally more expensive to maintain, they give developers a good reason to keep talented staff on board instead of laying people off between games. That alone can save time and money in hunting down staff.

It’s still risky

Just because The Witcher 3 fell off the chart quickly and Rocket League is still around doesn’t mean that games-as-a-service is free from risk. Live-service games are still subject to the hit-based nature of this medium.

Ubisoft’s The Division is an example. It is a blockbuster-scale shooter that the publisher designed to operate as a platform. But it also came and went in March 2016.

And while the consistent success of Warframe, Rocket League, and Ark are a positive sign for the viability of live-service games, few new games broke into that top 15 list on Steam since 2015. So it’s not like this is a recipe for a guaranteed hit.

But the industry isn’t turning to live-service games because they believe it guarantees them Siege like player numbers. It’s just that this model mitigates some of the risks.

Anthem vs. Andromeda

Electronic Arts may reveal over the next year how the live-service model is a smart evolution of how to make and sell games. The company has had two critically disappointing games from developer BioWare with this year’s Anthem and 2017’s Mass Effect: Andromeda.

Mass Effect: Andromeda is a largely single-player role-playing adventure. Like other games of that type, it came and went shortly after its debut. But many fans didn’t love it, and EA and BioWare decided to abandon the game and its plans for expansion-style downloadable content.

Now, two years later, Anthem is in a similar position. It is dealing with middling reviews and a mixed reaction from fans.

But unlike Andromeda, Anthem already has a roadmap of content and updates for the next 12 months. EA and BioWare have a clear path in front of them to turn Anthem into something that grows over time like Rainbow Six: Siege.

Again, nothing is guaranteed. BioWare still has to do the work to get Anthem into shape. But the point is that the live-service model gives it the chance to do that. The older model of building a standalone game and then adding onto that with DLC did not.

From a publisher’s point of view, it is far less risky to spend years working on a live-service game that you can eventually improve than on a product that bombs and everyone forgets about a few months later.